/MinMax-MTSP

A Reinforcement Learning Approach for Optimizing Multiple Traveling Salesman Problems over Graphs

Primary LanguagePython

MinMax-MTSP

Implementation of "A Reinforcement Learning Approach for Optimizing Multiple Traveling Salesman Problems over Graphs", which is accepted at "Knowledge-based System". If this code is useful for your work, please cite our paper:

	@article{hu2020reinforcement,
	  title={A reinforcement learning approach for optimizing multiple traveling salesman problems over graphs},
	  author={Hu, Yujiao and Yao, Yuan and Lee, Wee Sun},
	  journal={Knowledge-Based Systems},
	  pages={106244},
	  year={2020},
	  publisher={Elsevier}
	}

Dependencies

  • Python >= 3.6
  • Numpy
  • Google OR-Tools
  • Gurobi
  • PyTorch
  • tqdm
  • TensorboardX

Usage

Generating Data

Codes under gurobi and ortools can genearte data computed by Gurobi and Google OR-Tools respectively.

You are expected to generate validate/test data before running the training codes.

Training Model

The training codes are given under partition